We use web maps every day and never question why there is no coordinate system reference, yet our paper maps have more coordinate system information than the page can handle. Is it about time we reviewed our presentation of the map?
We [Geographers] create amazing worlds to be explored and yet when we put it on paper we feel the need to fill an eigth of the page with detail on the exact projected coordinate system used. Is this overkill or justified?
I am undecided. For years I have toiled and presented legal maps with every detail down to the false eastings & northings because it seemed to be the professional thing to do. It was the way I learned to present legal mapping and it “feels” right.
The past few years has seen more use of web mapping and online provision of geospatial information, one thing that is never present though, is the coordinate system used. What is more surprising is that no-one ever asks. It is automatically assumed that it is all correct. Here are a few examples: Google Maps,Bing Maps, or even the British Geological Survey…The point here is that there might be coordinates, but in what coordinate system? What is the transformation? What is the level of accuracy?
I’m now in a quandry and questioning the best practice moving forward. The logical answer is to provide the Coordinate Reference System or SRS/EPSG (delete as necessary) code to the online web map so that any allusion to location can be derived from the coordiante system information. What about paper maps/PDF outpu though? Do we NEED all that junk on the page if we are referencing the CRS/SRS/EPSG code? Surely by moving to this protocol we would be providing more space for our works of art…I mean maps.
Please vote you feelings on the poll below and I am happy to respond to any comment – Nick D
Thank you to all of you who commented and discussed my last QGIS v ArcGIS article, many of you pointed at some of the perspective taken with the article, for example, there was mention that it wasn’t fair as ArcServer wasn’t tested or that the 3D tools weren’t compared. Yes, you are correct, it would be unfair to compare tool to tool. QGIS is far superior in some areas and ArcGIS superior in others, but aren’t those the reasons why we choose the software?
This comparison is based on basic “GIS” functionality, the best way to see it is, “what would my 8yr old daughter use? “ it is easy to get too technical and start tearing apart the extensions or plug-ins but in an everyday use situation our lives are going to be wasted in the amount of time it takes to export a map or load a file.
This time around I will be comparing QGIS 2.8 (on windows) and ArcGIS 10.3 (on windows) and new boy on the block ArcGIS pro. These software were tested on identical Dell Xeon E5 1620 CPU workstations with 32GB of RAM and using Windows 7 Enterprise with no customisation and on fresh installs.
Price hasn’t changed for either product, QGIS is still open source and ArcGIS is still propreitary, for more information see the original article.
2. New Features
New in QGIS 2.8:
Advanced Digitising – Provides tools like parallel, perpendicular, locked angles etc.
Map frame rotation – Not just limited to the Layout/Composer
Embedded Heat Map Style
Group Layer Styles (QLR) – Save groups of layers and their styles
PostGIS performance improvements – BIG improvements!
Updated legend styling – More options for expressions & display
New function editor – Create your OWN functions!
New in ArcGIS 10.3
Buffer tool – Now has planar AND geodesic options
Spatial Join – Now has geodesic options
Conversion Toolbox has “From PDF” – Currently only supports from PDF to Tiff
Data management Toolbox, Create Raster Type
Data management Toolbox, Diagnose Version Metadata
Server gets mxd to WebMap – Although this listing is “like for like” I had to add this in…it is pretty cool and a nod to how far we have come in the industry.
Between the last versions of both software there have been no interface changes, with the exception of the introduction of ArcGIS Pro (This needs discussion separately). For more information about the 2 interfaces, see the original article.
3. Testing – Load Time
As this is a comparison of the 2 new updates, rather that witter on about my personal views, I thought it better to provide some test results. To make the test equal, the method I use was to start up the software being tested, leave it for 10 minutes (to ensure all elements have loaded) and the add ALL the EFH shapefiles & 2010 AIS from a folder in the root – C:GIS. If you wish to test the data yourself, it can be downloaded from here. The timer was started from the moment that the data is added to the data frame (through the add data button). The timer is stopped when all data has finished loaded, indicated in ArcGIS by the globe in the bottom right not spinning and in QGIS by watching the windows processes & seeing when the CPU demand dropped to zero again.
The time given is the average of the 5 runs which were made.
ArcGIS 10.2 = 27.27 seconds
ArcGIS 10.3 = 38.5 seconds
QGIS 2.6 = 9.103 seconds
QGIS 2.8 = 8.201 seconds
On average QGIS was 20 seconds faster to load the data, furthermore, there was a slight speed increase for QGIS from 2.6 to the newer 2.8
4. Testing – Analysis
What uses is your software if you spend your days waiting for it to finish a process? One test which I always run is a viewshed test. I put a single point down, in this test, a shapefile at 526451, 78576 (OSGB36 BNG). Then, using standard settings, I run a viewshed (ensuring the output raster resolutions are identical)
ArcGIS 10.2 = 42 minutes
ArcGIS 10.3 = 4 failed attempts & gave up*
QGIS 2.6 = 58 minutes
QGIS 2.8 = 1hr 16min
*To eradicate any issues with the data input or user error, the ArcGIS 10.2 map was saved and then opened in 10.3, the exact same parameters were used with no success.
5. Testing – Map Export
I seem to lose days exporting PDF files from GIS, so for me, this test is one of the most relevant, so that I can compare the amount of time consumed by the simple “PDF Export”.
For this test, the same EFH shapefiles used previously were loaded, no styles applied, then the bathymetric contours (also from the Marine Cadastre website) and the ESRI world countries shapefile. I then opened the layout/composer & set the orientation to landscape & the size to A3. Scale was set to 1:50000,000 then centered on the EFH data.
Once loaded and left to settle of 10mins, I set the PDF export to 500dpi and then made sure that both the QGIS & ArcGIS settings were the same (no layers, no georeferencing etc). This is the average times for 5 runs.
The results were as follows:
ArcGIS 10.2 = 3mins 18s – File size 795MB
ArcGIS 10.3= 3mins 30s – File size 903MB
QGIS 2.6 = 37.4s – File size 72MB
QGIS 2.8 = 35.5s – File size 69MB
6. Review of Niggles & Life savers
Although not entirely consistent with the rest of the test, I think it useful to keep track of the “niggles” to see whether they get addressed, my aim is to keep this open to readers to provide their own and as long as they are sensible, I’ll add them to the list. I would like to keep this to a maximum of 10 items per software.
Note: With reference to point 5, although there is a metadata plugin, there is is not a metadata engine or facility in the core, to be a fair comparison, I therefore have to dismiss this, even though it is quite good.
Note: With reference to point 6, there are some great methods for table joins outside of the core of QGIS, as above this cannot be taken into consideration unfortunately.
As well as comparing the times above, I also run through both software by opening a list of tools which are used within the office on a regular basis to ensure that the software is safe to be used.
As stated previously, it would be unfair to compare the tools as some tools are unique or paid for plugins. Needless to say, I found no issues in my testing.
In this comparison, I think that, although ArcGIS 10.3 has released a HUGE number of new tools and updates, they are, in large for specialists. These tools are unlikely to get used on a regular basis like you & I.
QGIS has shown some muscle and has, in this testing, shown itsef much faster. Furthermore one of my “niggles” from the last article has been dealt with, the shadows for points and fonts. Although there are only a handful of updates & new features, the tools (for me) are more likely to be used.
Again, as before, there is no clear winner as they are both great systems and suited to different professionals in different way. For basic use or an introduction to the world of GIS, you wouldn’t go wrong whichever you choose.
Since posting this I have been made aware that QGIS “Niggle” number 4 isn’t an issue. I could argue that the interface doesn’t make it clear that multiple field categories are possible but to be a fair comparison I must say that the function is possible and comparable with ArcGIS.
My apologies to Nathan Woodrow who had to endure a few hours of my stupidity to get to the bottom of it.